python --读rtf文件(转excel格式)

python --读rtf文件(转excel格式)

读rtf文件

pip install pywin32

from win32com.client import Dispatch, DispatchEx

# word = Dispatch('Word.Application')     # 打开word应用程序
word = DispatchEx('Word.Application')  # 启动独立的进程
word.Visible = False        # 后台运行,不显示
word.DisplayAlerts = False  # 不警告

path = r'C:\Users\EDY\Desktop\GD0101-06.rtf' # 写绝对路径，相对路径会拨错
doc = word.Documents.Open(FileName=path, Encoding='gbk')

for para in doc.paragraphs:
    print(para.Range.Text)

doc.Close()
word.Quit()

转excel格式

import os
import win32com.client as win32

def save_as_xlsx(fname):
    excel = win32.gencache.EnsureDispatch('Excel.Application')
    wb = excel.Workbooks.Open(fname)

    wb.SaveAs(fname+"x", FileFormat = 51)    #FileFormat = 51 is for .xlsx extension
    wb.Close()                               #FileFormat = 56 is for .xls extension
    excel.Application.Quit()


if __name__ == "__main__":
    package = "E:\\你的xls所在的文件夹\\"
    files = os.listdir(package)
    for fname in files:
        if fname.endswith(".xls"):
        	print(fname + "正在进行格式转换，请稍后~")
            save_as_xlsx(package + fname)
            print(fname + "格式转换完成O(∩_∩)O哈哈~")
        else:
            print("跳过非xls文件："+fname)